Hi,
I have some feedback regarding the new weekly stats.
Thanks for including the number of active translators, that's helpful. And thanks a lot for providing the stats at all, of course.
I'm not convinced about showing the number of jobs instead of units, though. This makes it harder (technically impossible) to judge the actual volume of available work and we can't directly compare it to older weekly reports that used units.
And if the goal is to understand how likely we are to find work at a specific time, using collections would make more sense.
Say, for example, that we notice a lot of jobs tend to be available around 2 am. We might decide to stay up longer, only to find out that these were either tiny jobs or bundled in one or two big collections that were picked up immediately by other translators.
I think an ideal solution would be to show both units and collections somehow, and maybe even combine the weekly and daily chart. I understand why they have been separated and maybe the resolution options of the software you are using are limited, but having one vertical chart for the whole week would make the most sense to me. The thickness of the bars could be reduced, and not every hour needs to be labeled.
The graphs for units and collections could be shown next to each other or combined into one, with the latter expressed through a color/value gradient (taking color-blind people into consideration). But I guess in combination with the Standard/Pro distinction that might get a bit much. Perhaps there is a better way to combine them.
Edit: Another idea would be to release the data e.g. in form of an additional XML file, so everyone can do their own analysis.
15件のコメント
Hi Chris! 😊
Thanks for sharing your feedback about the new weekly stats. I will share your comments and suggestions to the team. 💚🧡
I fully agree that this new format is not particularly legible or helpful. I'd much rather see the number of jobs AND the number of units, or only the number of units like before. There being 250 jobs doesn't mean much if they are all one word jobs, for example. Since a lot of jobs are very short, it's not a great way of knowing what the actual work volume is. It is an interesting stat, obviously, but only in conjunction with other stats such as the number of units, time of day, etc. Anyway, these were only my two cents on the topic. :)
Thanks a lot, Rica!
Another idea would be to indicate which jobs or collections were initially reserved for preferred translators, meaning they were likely not really available in the first place. But that might be information overkill and could also lead to discontent.
I have the feeling that this new system is designed to make it look like there is more work than there really is, because the only jobs that have appeared in my language pair (EN < SP) over the last three or four months have been for pennies. It's not that I haven't been quick enough to grab bigger assignments, it's that there literally haven't been any. Unless, of course, the siphoning problem is a reality (still waiting for news on this from the technical team) and the jobs never make it to either the feeder or the dashboard.
I believe the weekly stats should be expressed in a metric that's as close to the number of words as possible (words, segments, jobs, rather than collections).
This stats topic and the suspicion about 'siphoning' has just triggered me to write the following long rant, so be warned :-)
I don't believe that jobs are being snatched by a bot or something. After all, the strict time limits mean that you better start immediately in order to complete the collection on time. There is some security buffer, but the system definitely doesn't let you take a half-day break before you actually start. So what would you do with 20 big collections that a bot has 'bunkered' for you? Do "service flipping" and rent them out to other people for a lower rate (an almost criminal idea brought up by certain people on Youtube some time ago)? Machine-translate them and let the bot paste that output directly into the Gengo editor? Maybe I'm just naïve... but I believe that just wouldn't work. It seems too complicated to pull off on many levels, and I do believe that Gengo's system is robust enough to avoid this kind of cheating.
What I do see is that there has always been (at least since 2020, when I started here to fight the lockdown boredom) a low volume of jobs relative to the number of translators. That's why jobs are picked up so incredibly fast. There will always be jobs that you don't get to see because the website and RSS updates are only sent in specific time intervals. I think the list on the website only updates once per minute or so for each user - which is enough for others to snatch jobs that fly right under your radar. Maybe the RSS is different now, but when I tried using it, it would only update once per minute as well.
Up to a point, having many translators for few jobs is one of the core principles of how Gengo works: this kind of ratio (more or less) guarantees speed and the ability to absorb sudden peaks like the one we saw last week in EN>DE. Concerning the job volume as such, has anyone ever looked at the weekly numbers to draw a long-term trend chart? I wouldn't be surprised if the tendency was negative, simply because of the changes in the translation market, where some people think that for the simpler jobs it is enough to rely directly on the free version of DeepL. I'm on other, similar platforms where there has hardly been any work at all over the last few weeks, up to a point where it seems they are silently being phased out. One of them keeps showing jobs even when they are already being worked on, and still the listing is empty most of the time. Here on Gengo, at least there is something going on.. sometimes :-)
Besides all that, a high percentage of collections I see has a heart symbol next to it, so if you aren't a "preferred translator", you just won't see them. For example, most of the peak workload mentioned above had those hearts all over the place. (A pity that many of those collections lacked the context or a glossary that would have guaranteed a coherent translation of the whole thing...)
What's pretty intransparent to me is how many clients have actually chosen me as a preferred translator, and based on what experience or information. Based on the hearts I get to see, my impression is that you don't necessarily need to have done any previous work for that client. Maybe clients can pick you just based on some stats? Or the heart might appear automatically based on the GoCheck score? Maybe the Team can explain how it works...
@Rup75 What you say about the feeder and the dashboard makes a lot of sense, but in that case, there wouldn't be so many of us protesting about the lack of work. There would be voices saying "hey, I haven't noticed anything". Or so I think. Besides, it has been a non-progressive drop in activity: the jobs have practically disappeared the moment they "fixed" the feeder, or so it seems to me.
As for the clients who choose you as their preferred translator, mine seem to have disappeared too, because I had quite a few, with fairly large jobs and an almost weekly frequency, and again, from one day to the next, poof! they've vanished.
If what Gengo wants is for a million translators to fight for a handful of jobs so that clients don't have to wait at all, this is going to be like the Hunger Games and what they will get is that the whole system will collapse because, if we don't make money on this platform, what are we going to stay for?
What you say about this step drop is indeed a bit weird, but if you look into old forum posts, you will find that people have *always* been complaining about the low volume of work. Obviously, people do leave or limit their activity when they see that other things are more worthwhile or profitable for them. I have just returned from an almost year-long break myself...
- Concerning the weekly stats, why not just add the new stats (number of jobs and active translators) to the previous stats (number of units)? Technically speaking, I guess that shouldn't be very complicated.
- By the way, in my language pair, we are 3 active translators, and there is clearly no siphoning problem (the jobs are scarce, rarely picked immediately, and my browser's feed reader - not always activated - "sees" almost all of those reported in the weekly stats). I simply observe a gradual decline in jobs, which has led to a progressive reduction of the team (we were around 10 in 2020). I think this is mainly due to the changes in the translation market and Gengo has not recruited anyone in my pair since September 2021.
In EN>DE we are about 60 active translators per week, for about 900 collections, that would be about 15 per translator on average. But many (most?) of the collections are below 3 USD or even below 0.50, so you get the picture.
In theory, someone could do a trend analysis for their language pair based on the weekly overviews that still used units. Unfortunately, I didn't get those mails before the end of 2020, so well into Covid. If I recall correctly, they went to a different email account or I had been unsubscribed for some reason. At a glance, I didn't see any dramatic recent drop for EN>DE. Seems like there was more fluctuation to the positive side in the beginning of the mentioned timeframe, though, and now we have more consistent numbers at the lower end without those very good weeks. But overall, it looks pretty stable for the last one or two years. Maybe someone still has access to older weekly overviews from before Covid.
From a personal perspective, I started to have less time for Gengo right around the time when Covid hit, so it's hard to gauge if I would have earned less if I continued to work the same way. I feel like the beginning of Covid brought in quite a bit of jobs talking about Covid itself, and then the overall volume dropped a bit.
Regarding preferred translators, I think some customers are assigned to translators by Gengo/Lionbridge project managers, based on translator score or speciality. Not sure if I heard about that or if it's just an assumption. Theoretically, there might also have been a shift from using the Gengo platform to working directly with translators where it's more practical. (Note that this is complete speculation and that there are sufficient other explanations for the decline in jobs, though.) There used to be, maybe still is, a kind of separate projects platform, but it doesn't seem to be used much, perhaps not at all, these days.
Hello, I have got a question concerning stats: what exactly does "active" translators mean? The ones who actually check their feeds and saw the jobs from the stats? Or all translators for that pair, even if they don't use the platform anymore but haven't deleted their accounts? I'm asking because in week 1 there were 10 active translators and in week 2 15 - does that simply mean there have been 5 new accounts in one week? If this is the case, why does Gengo accept new translators in a pair with next to no jobs for weeks and months?
Good question. I had automatically assumed that an active translator is a translator who has translated at least 1 collection during that week. So it can go up and down every week even if there aren't any new translators.
That is not possible, as there has been one (1) job in that week, but 15 active translators....
Hi everyone. 😊
I apologize for the late response to this thread. I will consolidate all the additional comments you've shared here (and in the forms) and give them to the team. We are grateful for these insights that will help us improve this initiative.
As some partners mentioned in the other threads, there is a decline in the volume of work, and the dynamics of the industry are continuously changing. We assure you that our teams are working on it to create more opportunities for our partners.
As we don't have any control over the changes in the customers' needs and demands, we started cleaning up our pool of translators to lessen the competition. Even though it's not much, we are doing our best to tackle the issues our partners are facing.
To answer the question about the active translators data, it is the number of translators who have been online on the platform in the previous week, regardless of whether they work on a job or not.
Thanks a lot, Rica and devs!
I see that the current weekly overview already uses collections, which is better than jobs. I hope we'll still see units being incorporated again in some way in the future.
Could you please clarify what cleaning up the pool of translators looks like in practical terms? E.g. is this about combatting abuse of the system or revoking qualifications at the lower end?